Colloidal Structure of Rubber in Solution
EFFECTS OF PRECIPITANTS’
S. D. GEHMAN AND J. E. FIELD The Goodyear Tire & Rubber Company, Akron, Ohio
R
UBBER solutions are somewhat unstable and their
colloidal structure, usually characterized by Viscosity, can be influenced by a large number of agenciessuch BS light, oxygen, and small amounts of various reagents. The colloidal changes caused by the addition to a rubher solution of typical nonsolvente or precipitants, such aa alcohol or acetone, are distinguished by the fact that they are reversible. If the precipitant is removed, the original solution can be recovered. The reversible colloidal changes, which result eventually in precipitation, were studied in the work here reported as a means for acquiring knowledge of the colloidal structure of ruhher in solution. As in a former article (8) an effort is made to utilize comparatively recent advances in the general theory of the liquid state (3, 6, 9). Previous publications dealing with the effects of nonsolvents on rubber solutions have been predominantly of technical interest and limited almost exclusively to viscosity measurements. LeBlanc and Kfoger (f.S), Messenger and Porritt (g, 16), Kawamura and Tanaka ( I f ) described the 1
For the 6rat p8per in this sedea. BBB literature citation (8)
The colloidal changes caused by the addition of precipitants to rubber solutions have been observed by viscosity measurements and measurements of the intensity and depolarization of the transversely scattered light, as a means of securing information about the colloidal structure of rubber solutions. The light scattering measurements show that colloidal changes occur which are not apparent from the viscosity data. This is explained by the fact that the light scattering depends upon the geometry of the structure and the viscosity upon the potential energy in the structure. The results are considered to be further evidence that, in rubber solutions, the rubber molecules exist in clusters or groups, the size, shape, and “interlocking” of which depend upon the molecular forces in the solution.
viscosity changes which ensue upon the addition of various nomofvents to rubber solutions. Whitby ($6) classified the reagents which can he used to affect the viscosity of rubber solutions. In a later publication ($3) he discussed an interesting experiment during which incipient precipitation was ohserved with the ultranlicroscope. Lens (1.6) reported the results of a few measurements of the effect of alcohols on the relative intensity of the light transversely scattered by benzene solutions of rubber. Ford (6) and Fabritziev, Batiko, and Pakhomova (4) studied the modification of technical rubber cements by the use of nonsolvents. The colloidal effects of precipitants on solutions of high polymers such as polystyrene have been observed and are of interest in connection with this work ( I S , $1, 82,26).
Apparatus and Methods
FIGURS1. SEPARATION OF SOLAND GELRUBBER nY DLEWSION 1031
This study of the colloidal changes caused by the addition of precipitants to rubber sohtions was carried out by nieans of viscosity measurements and measurements of the intensity and depolarization of the transversely
INDUSTRIAL -4ND ENGINEERING CHEMISTRY
1032 I5
I
I
I
1
VOL. 30, NO. 9
port of Midgley and Henne’s e x p e r i m e n t where some abnormal disintegration or oxidation of the rubber seems to have occurred.
16 L4 A2
Viscosity Effects 240
The addition of a precipitant, such as methyl alcohol, to a rubber solu3 6 9 5 35 tion, even in a much 160 smaller amount than is 2 9 required for precipita30 t i o n , causes such a marked decrease in the viscosity that a pro25 found change in the colloidal structure may readily be suspected. 80 Figure 2 gives typical curves illustrating how the viscosity of a rubber solution is altered by the addition of a precipitant. For the curves of Fig, I I /o ures 2 and 3, in contrast 5 /O /s 2 yo PR€C/P/TANT to most of those in the FIGURE2. VISCOSITY CHANGES CAUSEDBY l i t e r a t u r e , the rubber ADDITION OF PRECIPITANTS concentration was kept constant as precipitant was added. This was scattered light, employing the appadone by the addition of ratus, technic, and theoretical analysis an appropriate volume which were previously described in deof more concentrated tail (8). .2 4 6 8 solution. The concenPale crepe rubber purified by acetone R ALCOHOL tration of the precipiextraction and diffusionin petroleum ether FIGURE3. EFFECT ON VISCOSITYCURVES tant, p1otted:as abscissa, was used (1). Midgley and Henne (IS) OF ( A ) LARGEVARIATION IN CONCENTRATION is the yolume of preAND (B) TEMPEIRATURE recently criticized the process of diffusion cipitant present divided for the seDaration of sol and gel rubber. by the volume of the basing thilr conclusions on thcresults of solution after the precipitant is added, multiplied by 100. an experiment in which ethyl ether was the solvent. The viscosities were measured a t 30” C. The solutions The writers’ experience has been that petroleum ether is a of Figures 2 and 3 were exposed to air during the measmore reliable medium than ethyl ether for effecting a separaurements, but a control solution showed a negligible change tion of sol and gel rubber. With ethyl ether a disintegration in viscosity during the period of time taken for the measand dispersion of the gel rubber sometimes occur. In urements. several controlled experiments using two different brands of The effects of the precipitants upon the viscosity have been technical absolute ether, it was possible to show that with one generally attributed to desolvation of the rubber, usually of them the disintegration always occurred whereas with the thought of as existing in the solutions in the form of swollen other it did not. The cause of this disintegration would, micelles. The authors wish to bring a different point of therefore, appear to be some impurity which may be present view to the explanation of these curves. In the previous in ethyl ether (24). The difference in the diffusion process is work (8) it was deduced from the light scattering measureillustrated in Figure 1. Each liter bottle contained 15 ments that the rubber molecules formed clusters in solution, grams of acetone-extracted pale crepe rubber. Bottle A was the size and shape of which depended upon the solvent and filled with the ethyl ether which caused disintegration of the the concentration. One of the fundamental differences begel rubber. In bottle B, which was aIso filled with ethyl tween such clusters and a colloidal micelle is that each cluster ether, the diffusion proceeded in the normal manner. Bottle represents an equilibrium condition under the acting molecuC was filled with petroleum ether. The photograph was lar forces, whereas the colloidal micelle is a discrete entity taken after the bottles had stood in the refrigerator for 2 which maintains its identity in different solvents and in the months. The gel rubber in C retained more strength than solid rubber. A more accurate description of the clusters in B, as is apparent from its greater resistance to settling. would probably represent them as having a statistical exHowever, the sol rubber from B and C, redissolved in purified i s t e n c e i . e., as being continuously broken up by the thermal ethyl ether, gave solutions for which the intensity of the agitation and reforming to an average size. The decreased scattered light was practically identical. From this and thermal agitation would help to account for the larger size of similar experiments, as well as from the other evidence availthe clusters a t higher concentrations. able in the literature (1.2,20), the writers believe that it is The viscosity of a rubber solution depends not only upon possible to effect a much better separation of the sol and gel the size and shape of the clusters but also upon the forces rubber by the process of diffusion than is implied in the refl€THYL .4LCOHOL I
6
zoo
’
~
SEPTEMBER, 1938
INDUSTRIAL AND ENGINEERING CHEMISTRY
between them. It is inconceivable that the molecular forces between the rubber molecules should cause them to gather into such loose clusters and yet that the forces between the clusters could be ignored. Evidence for this dependence of the viscosity upon the molecular forces was given by showing that the dielectric molecular polarization of a series of solvents was closely correlated with the specific viscosities of the rubber solutions. The dielectric molecular polarization in itself cannot account for all the viscosity phenomena. For this purpose, a more exact mathematical representation of the molecular forces involved would be required. As for solvation, .there is no logical reason for assuming that it exists to any greater or lesser degree for long molecules than for
-~~
~
0.2 0.3 0.4 0.5 RUBBER CONCENTRAT/ON /N E T H E R O./
p
FIGURE4. RELATIONBETWEEN RUBBER CONCENTRATION AND REDUCTION OF VISCOSITY
short ones, so that its role in the viscosity phenomena for rubber solutions is, in most cases, probably of minor importance. The addition of methyl or ethyl alcohol to a rubber solution, as shown in Figure 2, causes an abrupt decrease in the viscosity at first. The dielectric molecular polarization of the alcohols is high compared to that of benzene-52.1 for ethyl alcohol, 36.8 for methyl alcohol, and 26.3 for benzene. A viscosity decrease is therefore expected. For ethyl ether the dielectric molecular polarization is 54.5, so that the effects of the alcohols on the viscosity of this solvent are not so obvious. However, it is impossible to calculate the dielectric molecular polarization of such a mixture in an additive way. Thus, it has been shown (IO) that the dielectric molecular polarization for an ethyl ether-ethyl alcohol
qo
1033
mixture has a maximum value greater than that of either component a t about 30 per cent alcohol. It is perhaps with such a maximum that the abrupt change in slope of the viscosity curves upon further addition of precipitant is connected. For a given solvent and precipitant, the position of the break in the viscosity curve is very insensitive to the rubber concentration, as shown in Figure 3A, and also to the rubber plasticity (4). This is evidence that the same forces are primarily responsible for the viscosity of these rubber concentrations. The extent of the drop in viscosity a t the "break" depends upon both the rubber concentration and the plasticity of the rubber. This dependence upon concentration is apparent in Figure 3A and also in Figure 4. Figure 4 shows the per cent change in specific viscosity caused by the addition of one per cent of ethyl alcohol to rubber solutions of various concentrations in ethyl ether. The viscosities were measured in the sealed viscometer so that the solutions were not exposed to air at any time. The alcohol was contained in a sealed evacuated tube which was broken in the viscometer. The solvation would certainly be expected to be greater at the lower rubber concentrations so that these results cannot be reconciled with any interpretation based on desolvation as the cause for the decrease in viscosity. Figure 3B shows that the effect of the temperature on the specific viscosity is rather small even when a precipitant is present. This results from the fact that the absolute viscosities of the solution and the solvent have the same functional relation with the temperature; therefore their ratio, from which the specific viscosity is calculated, is not so much affected. This again is evidence for the existence of EL molecular structure in the rubber solution similar to that now identified with the liquid state. For some rubber solutions the specific viscosity actually shows a small increase with increase in temperature. The changes in specific viscosity of the rubber solutions due to temperature do not appear to be explicable on the basis of changes in the dielectric molecular polarization of the solvents. A comprehensive and precise investigation of the viscosity changes due to precipitants in terms of all the variables in-
ALCOHOL
% ACTi-ONE
FIQURB5. RELATIVE INTENSITY OF LIGHTSCATTERED BY ( A ) SOLUTIONS CONTAINING ALCOHOL AND ( B ) SOLUTIONS CONTAINING ACETONE
1034
INDUSTRIAL AND ENGINEERING CHEMISTRY
VOL. 30, NO. 9
ber concentration was kept constant by addition of the required volume of a more concentrated solution, the dip was absent and the curves showed a very gradual rise from the start. Figure 6 shows that the addition of alcohol to a heptane solution causes an initial drop in the intensity of the scattered light which is followed by a steep rise, similar to the curves of Figure 5A.
70
ALCOHOL
FIGURE 6. INTNNSITY MEASUREMENTS FOR HEPTANE SOLUTION
A
volved, especially in relation to the dielectric constants and molecular polarizations of the solvent and precipitant mixture, should be of considerable interest for the study of the colloidal structure of rubber in solution. This work, however, was limited to the study of the colloidal changes occurring in a few typical cases; measurements of the intensity and depolarization of the transversely scattered light were largely utilized.
Intensity of Scattered Light % RUBBER CONCENTRAT/ON /N EPf"R
The measurements illustrated in the curves were carried FIGURE 7. LIGHT INTENSITY-RUBBER CONout on solutions which were exposed to air for the addition of CENTRATION CURVES FOR SOLUTIONS CONTAINthe precipitants. However, enough measurements were ING ALCOHOL made in sealed evacuated cells, where the precipitant was added by breaking a sealed tube, to ensure that the results Figure 7 demonstrates the marked differences in the curves here given were not significantly affected by the exposure. of rubber concentration vs. intensity of scattered light when The use of ethyl ether and heptane solutions, for which the intensity of the scattered light is fairly high, rendered dust the solvent contains different percentages of alcohol. For 11 per cent alcohol in the ethyl ether solutions, the curve is contamination less important than i t would have been for linear, similar to the curve that would be obtained for a solvents of higher refractive index. The solvents and prelyophobic colloid. When no alcohol is present, the curve cipitants used had approximately the same refractive indices. is concave towards the rubber concentration axis and beFigure 5A shows the effect of alcohol on the relative incomes very flat a t higher rubber concentrations. For 5 tensity of the scattered light for ethyl ether solutions of sol per cent alcohol the curvature is intermediate between these rubber a t several concentrations. For comparison, the extremes. straight line in the figure shows the change in intensity for the 0.5 Der cent rubber solution due to dilhion with the corresponding volume of solvent. The addition of the first trace of alcohol causes a sharp drop in the intensity. -8 O0.06 .O At a concentration somewhat less than that of the break in theviscosity curve, the intensity curve becomes flat; then a t higher concentrations of alcohol it rises rapidly, a change not shown by the O.OZ 0.04 viscosity curves. Figure 5B gives the changes in 3.02 the relative intensity of the scattered light caused by the addition 0.08 0./o of acetone to ethyl ether solutions of sol rubber. In contrast to the 3.08 0.06 curves of Figure 5.4, the curves for acetone addition do not have 9.06 0.04 an abrupt initial drop. The slight dip in the curves can be explained 3.04 0.02 as being due to the dilution. This is apparent from the straight line representing the decrease in in30 /O 6 fZ gb AC€TON€ 2o %ALCOHOL tensity for the one per cent solution oaused by dilution with solFIGURE 8. PLOTOF DEPOLARIZATION FACTORS FOR ( A ) ALCOHOL ADDITION AND ( B ) vent. Furthermore, when the rubACETONE ADDITION
, Ilir-e44u*I
SEPTEMBER, 1938
INDUSTRIAL AND ENGINEERING CHEMISTRY
1035
+
Depolarization of Scattered Light Reference should be made to the earlier paper (8) for the significance of the depolarization factors. If J1 and JZ are the intensities of the vertical and horizontal components in the scattered light, the following table defines the factors: State of Polarization of Incident Beam
Depolarization Factor
Definition of Factor
Unpolarized Vertically polarized Horizontally polarized
Pu
JdJi
PW
Ji/Ji
Ph
Ji/Ja
cipitant. The anisotropy factor, 2 p , / ( l p,) (Figure 8 ) is practically constant until the precipitation point is approached, when it rises rapidly. The initial increase in size of the clusters is consistent with the initial decrease in scattered light intensity observed in Figures 5 A and 6. This decrease in intensity was not observed for acetone addition, however. Corresponding to the increase in the size of the light scattering units, there occurs a decrease in the specific viscosity. These facts may be qualitatively explained by considering that the first addition of precipitant influences the molecular forces in such a way that neighboring clusters
The interpretation of the depolarization measurements here given is in accordance with the theory developed by Krishnan. Mueller (17) recently suggested an alternative explanation which can be considered only tentative. Gans (7) criticized one of Krishnan’s formulas. The writers’ experimehtal results favor the formula of Krishnan rather than that of Gans. Gans’ formula gives consistently low values of p, as calculated from Ph and put which is not in accord with his idea that the failure of his formula to agree with experimental results is due to experimental errors in py. The curves of Figure SA illustrate the effect of alcohol addition on pu, p,,, and the anisotropy factor, 2p,/(l p,,), for a one per cent solution by volume of sol rubber in ethyl ether. Figure 8B gives the same curves for acetone addition. In Figure 9A are plotted the factors Ph and A p , = p, - m 2Pv P + t for alcohol addition, and in 933,the same factors for acetone addition. The form of the curves was verified for concentrations of rubber from 3 to 0.5 per cent. Figure 10 shows the results of light scattering measurements for the addition of acetone to a 1 per cent solution of sol rubber in heptane.
3.06
9.04
2 02
+
aoo
20
/O
Discussion of Results In attempting to correlate the viscosity and light scattering results, it is necessary to keep in mind that the light scattering phenomena are due to the geometrical distribution and orientation of the rubber molecules in the solution. The viscosity, on the other hand, depends in a sense upon the stability of the structure or, more precisely, on the energy required to displace a unit of the structure (19). In the earlier paper (8) it was shown that the colloidal units responsible for the light scattering are probably anisotropic clusters or groups of rubber molecules which are large compared to the wave length of light. Since an increase in A p , and a decrease in Ph correspond to an increase in size of the scattering unit, the curves of Figures 9 and 10 are consistent in indicating an increase in sizi of the clusters as the first effect of ;he addition of pre-
0.02
0 . 4 I--
FIGURE 9.
30
40
I
a
q o A CE TONE OF INTENSITY AND DEFIGURE 10. VARIATION POLARIZATION OF A HEPTANE SOLUTION
unite into larger units. This has the effect of diminishing the structure throughout the solution and causes the viscosity to fall. The structure of a rubber solution i s thus represented as having a twofold character. There is structure in the colloidal units or molecular clusters, and there are structural “interlocking” relations of these units throughout the solution. As the precipitant is added, the structure in the colloidal unit approaches that of solid rubber, and the structure in the solution is progressively diminished until eventually precipitation occurs and the structure completely vanishes. After the initial growth in size of the clusters, the light scattering measurements show further colloidal changes as DreciDitant is added. For alcohol addition to the ethyl ether solutions (Figure 5 A ) after the concentration of alcohol becomes approximately 3 per cent, the intensity curve rises continu0.01 ously until precipitation occurs. This indicates that a breaking down of the scattering units occurs. This decrease. 0.8 in size is again verified by the changes in p,, and A p , (Figures 9 and 10). pb reaches its highest value near the pre0.6 cipitation point. The value a t this point is higher than in the original solutions. The corresponding physiI I I /O 20 30 cal picture may be that the large clus3 6 9 /P 70 ACE2-ONE -% ALCOHOL ters become unstable a t higher concenAp, AND p i FOR ( A ) ALCOHOLADDITIONAND ( B ) ACETONEADDITION trations of precipitant and break up to .
0.03
cg
’
&
1036
INDUSTRIAL AND ENGINEERING CHEMISTRY
form a number of comparatively compact clusters with reduced solvation. This view is supported by the increase in the anisotropy factor. When precipitation actually begins, the ends of the curves of Figure 9 indicate the growth in particle size which results in settling out of the rubber. The light scattering measurements a t this point are sensitive to the temperature. From the results and the discussion, it is apparent that the colloidal changes which take place upon the addition of precipitants to rubber solutions may advantageously be considered from the same standpoint as other fundamental problems dealing with the molecular forces and structure in liquids and solutions. Furthermore, the description of the solutions which has been offered tends toward a reconciliation between the extreme views of the exponents of the molecular and the micellar colloids. At higher rubber concentrations, the molecular clusters are so large and stable that the solutions have a micellar character; at very low concentrations, these clusters are so small and so subjected to thermal agitation that the colloidal phenomena approximate those which would be predicted as a result of the presence of individual threadlike macromolecules.
Literature Cited (1) Caspari, W. A,, J . SOC.Chem. Ind., 32, 1041 (1913). (2) Dawson, T. R., and Porritt, B. D., “Rubber,” p. 67, Croydon, Rubber Research Assoo., 1935. (3) Debye, P., Chem. Rev., 19, 171 (1936). (4) Fabritziev, B. V., Buiko, G. N., and Pakhomova, E. A.. Rubber Chem. Tech., 9, 428 (1936).
VOL. 30, NO. 9
(5) Faraday SOC., “Structure and Molecular Forces,” London,
Gurney and Jackson, 1936. (6) Ford, T. F., IND. ENQ.CHEM.,28, 915 (1936). (7) Gans, R., Physik. Z . , 38, 625 (1937). (8) Gehman, S. D.. and Field, J. E.. IND.,ENG.C H ~ M . , 79.-1 , .29. (1937). (9) Herzfeld, K. F., J . Applied Phys., 8, 319 (1937). (10) Kanamaru, K., and Ueno, S., Kolloid-Z., 79, 77 (1937). (11) Kawamura, J., and Tanaka, K., Rubber Chem. Tech., 5, 626 (1932). (12) Kemp, A. R., IND.ENQ.CHEM.,30, 154 (1938). (13) LeBlanc, Max, and Kriiger, M., Kolloid-Z., 33, 168 (1923). (14) Lens, J., Rubber Chem. Tech., 6, 265 (1933). (15) Messenger, T. H., and Porritt, B. D., J . Research Assoc. Brit. Rubber Mfrs., 1, 7 (1932). (16) Midgley, T., and Henne, A. L., J . Am. Chem. Soc., 59, 706 (1937). (17) Mueller, H., Phvs. Rev., 52, 223 (1937). (18) Sohulz, G . V., 2. physik. Chem., 179A, 321 (1937). (19) Smallwood, H., J . Applied Phys., 8, 505 (1937). (20) Smith, W. H., and Saylor, C. P., J. Research Natl. Bur. Standards, 13, 453 (1934). (21) Staudinger, H., and Heuer, W., 2. physik. Chem., 171A, 129 (1934). (22) Staudinger, H., and Mojen, J. P., Kautschuk, 12, 159 (1936); Rubber Chem. Tech., 9, 579 (1936). (23) Whitby, G. S., Colloid Symposium Monograph, 4, 203 (1926). (24) Whitby, G. S., Trans. Inst. Rubber Ind., 6 , 40 (1930). (25) Whitby, G . S., and Jane, R. S., Colloid Symposium Monograph, 2, 16 (1924). (26) Whitby, G. S., MoNally, J. G., and Gallay, W., Ibid., 6, 225 (1928). RECEIVED April 5, 1938. Presented before the meeting of the Division of Rubber Chemistry of the American Chemical Society, Detroit, Mich., March 28 and 29, 1938.
Chemistry of Soft Rubber Reversion and Nonreversion in LowSulfur Compounds
Vulcanization’ B. s. GARVEY, JR., AND D.B. FORMAN The B. F. Goodrich Company, Akron, Ohio
4
CCUMULATION of evidence in recent years has led to a wide acceptance of the view that, chemically, vulcani-L zation is the establishment of cross bonds between the fiber molecules of unvulcanized rubber. The usual concept is that of a chemical cross bond typified by the sulfur bridge (4). It was recently proposed (2) that “mechanical cross bonds” might be formed by the interlocking of the molecules as they become kinked as a result of &-trans isomerization at some of the double bonds. It was also suggested that free rotation around single bonds might unkink the molecules. This straightening of the molecules would result in reversion of the vulcanized structure. This theory suggested that a study of reversion would be of value. For this purpose low-sulfur compounds are most suitable because the number of chemically stable sulfur bridges is small. Consideration of the experiments reported here suggests the interesting possibility that in vulcanized compounds there exists a sort of dynamic equilibrium between the formation of cross bonds and their destruction, which results in the maintenance of an adequate number of cross bonds although the individual cross bonds are not permanent.
The general methods of mixing, curing, and testing were previously described (3). Since PBA and Altax gave the most clear-cut distinctions between reverting and nonreverting types of acceleration, the complete data are given for these two accelerators. Similar tests were made with the other accelerators, but only the conclusions from them are reported here.
Reversion Tests The base recipe used was: First latex crepe Zinc oxide (lead-free) Sulfur Accelerator
100.0 8.0 0.5 2.5
The stocks were cured from 5 to 480 minutes a t 1 4 2 O C. (288’ F.). Table I gives test data for the compounds accelerated with Altax and PBA. 1
INQ
Five artioles in this series have appeared in INDUSTRIAL AND E N Q I X E ~ R CEZIVISTRY [25, 1042,1292 (1933): 26, 434,437 (1934); 29,208 (1937)l.
